Physically Accurate Lighting Simulation in Computer Graphics Software

A.B. Khodulev, E.A. Kopylov


Prev Contents    Introduction    Lightscape System    Specter System    Radiance System    Scenes for Comparison    Experimental Results    Conclusion    Acknowledgments Next

INTRODUCTION

In this paper we investigate capabilities for physically accurate global lighting simulation in several computer graphics systems. We compare the following systems:

It should be mentioned that we used restricted version of LVS distributed for educational institutions.

All these systems allow to do global lighting analysis but they use different approaches to solve the global light propagation equation ("rendering equation", [Kaj86]).

LVS uses radiosity method [CCWG88] for global light propagation simulation and backward ray tracing for rendering (image generation).

Specter uses bi-directional ray tracing with forward Monte Carlo ray tracing used for producing view-independent illuminance distribution in a model (the so called illumination maps) while in further invocations of backward ray tracing for creating high-quality view-dependent images information is retrieved from the illumination maps.

Radiance uses backward Monte Carlo ray tracing. It also stores illuminance distribution as a set of point samples (that is used instead of the Specter illumination maps).

As we are mainly interested in physical accuracy of the systems, the following criteria are used for comparison:

Sections 1-3 below analyze each system from the viewpoint of the above criteria with the exception of simulation accuracy that was examined experimentally. Results of this experimental comparison are presented on "Scenes for Comparison" and "Experimental Results" pages.


Prev Contents    Introduction    Lightscape System    Specter System    Radiance System    Scenes for Comparison    Experimental Results    Conclusion    Acknowledgments Next

Copyright 1996 Andrei B. Khodulev, Edward A. Kopylov.- All Rights Reserved [Home]